Sensitivity and identifiability of rheological parameters in debris flow modeling

Abstract

Abstract. Over the past decades, several numerical models have been developed to understand, simulate and predict debris flow events. Typically, these models simplify the complex interactions between water and solids using a single-phase approach and different rheological models to represent flow resistance. In this study, we perform a sensitivity analysis on the parameters of a debris flow numerical model (FLO-2D) for a suite of relevant variables (i.e., maximum flood area, maximum flow velocity, maximum height and deposit volume). Our aims are to (i) examine the degree of model overparameterization and (ii) assess the effectiveness of observational constraints to improve parameter identifiability. We use the Distributed Evaluation of Local Sensitivity Analysis (DELSA) method, which is a hybrid local–global technique. Specifically, we analyze two creeks in northern Chile (∼29∘ S, 70∘ W) that were affected by debris flows on 25 March 2015. Our results show that SD (surface detention) and β1 (a parameter related to viscosity) provide the largest sensitivities. Further, our results demonstrate that equifinality is present in FLO-2D and that the final deposited volume and maximum flood area contain considerable information to identify model parameters.

A modeling methodology to study the tributary-junction alluvial fan connectivity during a debris flow event

“…D' Agostino and Tecca (2006) suggest that SD works as a minimum physically plausible flow depth. In addition, Zegers et al (2020) reported that SD is one of the two more sensitive parameters in the model and, even though it is not part of the rheological model, it can surrogate the flow rheology.…”

“…is presented in Zegers et al (2020) and porosity p is set equal to 0.3 for facies F 1 and F 2 (Nicolleti and Sorriso-Valvo, 1991).…”

“…Thus, to reduce over parameterization and in the calibration process). Parameters α 1 and α 2 were fixed since they are the least sensitive parameters of the model (Zegers et al, 2020) and were set equal to 0.0075 poises and 0.152 dynes cm −2 , respectively, using the values obtained by Zegers et al (2017). Conversely, the SD parameter has a considerable influence on the results (Zegers et al, 2020).…”

“…Parameters α 1 and α 2 were fixed since they are the least sensitive parameters of the model (Zegers et al, 2020) and were set equal to 0.0075 poises and 0.152 dynes cm −2 , respectively, using the values obtained by Zegers et al (2017). Conversely, the SD parameter has a considerable influence on the results (Zegers et al, 2020). This parameter sets a minimum physically plausible flow depth, allowing the mixture to stop on the alluvial fan (i.e., the mudflow stops if the local flow depth is lower than SD).…”

“…We used the algorithm of Zegers et al (2020) with slight modifications to create a calibration routine that allows us to run the model N times and automatically apply screenings to determine the model's best fit. Similar to Mergili et al (2018), we assigned the pixels within the observed deposit as observed positives (OP) and pixels outside the deposit area as observed negative (ON).…”